genetic counsellingprematurely, andonehaemophilic maleandonenormalmaledid not survive fetoscopy....

Journal of Medical Genetics, 1982, 19, 26-34

Genetic counselling in haemophilia by discriminantanalysis 1975-1980*EMILY S BARROWt, CONNIE H MILLER+, HOWARD M REISNERt, ANDJOHN B GRAHAMt

From t the Department ofPathology, Curriculum in Genetics, and Center for Thrombosis and Hemostasis,School of Medicine, University of North Carolina, Chapel Hill, North Carolina; and +Division ofPediatricHematology, Cornell University Medical Center, New York, USA

SUMMARY Between January 1975 and January 1980 we counselled 214 possible or obligatorycarriers of haemophilia A in an attempt to provide the counsel they needed and to determine theutility of a probabilistic method of assigning the heterozygous genotype (carrier detection). Wefound the method of assignment to be quick and easy to use, and the single output (the final prob-ability favouring carriership P(C)) to be understood by most counsellees. The final probabilitiesobtained were either very high or very low in 80% of the women, which allowed us to give clearcounsel in four instances in five. The final probabilities could also be used to relate Mendelianexpectation to observation within each of three subsets of women (18 mothers of sporadic haemo-philiacs, 78 sisters, and 62 more distant relatives), while the aberrant likelihood ratios of 6/36 (17%.)of the obligatory carriers provided an estimate of the false negative diagnostic rate owing to lyonis-ation. There was no significant age effect on VIII:C or VIIR:Ag levels of obligatory carriers, andthe VIII:C levels of the obligatory carriers who had received the gene from their fathers did notdiffer from those of the obligatory carriers who had received the gene from their mothers. The ratiosof high:low probabilities among the sisters and distant relatives of haemophiliacs conformed togenetic expectation, while the ratio among mothers of sporadic haemophiliacs suggested that theirexpectations of carriership were greater than 80%.Twenty of the 214 counsellees (9%) were pregnant on the first visit, and 13 of those with low

P(C)s (0 00-33) went to term and delivered 11 non-haemophilic children. Four with P(C)s between0 50 and 1 00 requested amniocenteses, and one male was aborted. Three who were obligatorycarriers also requested amniocentesis which led to the abortion of a second male.

Seven women who were assessed before pregnancy and found to have high P(C)s returned afterbecoming pregnant. All requested amniocentesis (one twice) and fetoscopy was requested for six ofthe seven males discovered, one male having been aborted before fetoscopy became available. Ofthe six males who were fetoscoped, three normal males reached term, one normal male was bornprematurely, and one haemophilic male and one normal male did not survive fetoscopy.

The discovery of sex limited (now X linked) inherit- counselling using discriminant analysis based onance by Morgan in 1910 provided the Mendelian factor VIII coagulant assays has been a possibilitybasis for genetic counselling in haemophilia A.' It is for many years,3 it was a scientific breakthrough innot clear when physicians became widely conscious laboratory studies of factor VIII in the early 1970sthat inheritance of eye colour in Drosophila and which rekindled interest in carrier detection. Almosthaemophilia in humans followed the same pattern, simultaneously, Bennett and Huehns in the UK4 andbut Macklin may have been the first to specify the Zimmerman et alin the USA56 discovered that thereoutcomes expected from the various haemophilia- was a disproportion between the levels of coagulantrelated matings in a medical journal.2 While genetic factor VIII (VIII :C) and an antigen related to factor

VIII (VIIIR:Ag) in the plasma of haemophiliacs.*Supported (in part) by research grants (HL.06350, HL-24207, and They also observed a simnilar, though smaller,HL-23698) from the National Institutes of Health. disproportion in the plasma of many obligatoryReceived for publication 9 April 1981

26

on August 7, 2020 by guest. P

rotected by copyright.http://jm

g.bmj.com

/J M

ed Genet: first published as 10.1136/jm

g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/

Genetic counselling in haemophilia by discriminant analysis 1975-1980

carriers. The basic finding was quickly confirmed,7-'0and the feeling became widespread that the problemof haemophilia carrier detection had been solved.The original wave of enthusiasm was succeeded bythe inevitable reaction, workers who agreed inprinciple but pointed to inherent difficulties andsuggested improvements."1'-6 The intensity of thedialogue in the mid-1970s attracted wide attentionamong haematologists and led to two importantdevelopments: (1) a prospective evaluation of themethod,17 and (2) establishment of a committee bythe WHO to give considered advice.18 The last twostudies seem to have provided most of the infor-mation needed, and there have been few originalpapers on the subject since 1977.19 20 The back-ground, principles, and methods of carrieridentification in the haemophilias were summarisedin a comprehensive article in 1979.21The purpose of this paper is to describe the

application of the state-of-the-art procedure duringa period of 6 years and to evaluate its use.

Materials and methods

SUBJECTSThe 214 female counsellees seen between January1975 and January 1980 were related in a variety ofways to approximately 600 men with haemophilia Awho had been seen at our institution. Those notpregnant on the first visit included 36 obligatorycarriers, 78 sisters of haemophiliacs, 62 second andthird degree relatives, and 18 mothers of sporadichaemophiliacs. Twenty additional women at riskwere seen after becoming pregnant. (Sixteencounsellees from haemophilia B kindreds and 12from kindreds with other disorders of haemostasiswere also seen during the 6-year period.) All subjectswere interviewed by at least two of us, and strongefforts were made to assure a return visit. The

|Hoeoph@sa|| 6eetic |Personal ||Famitnc physicion

;p fy L

1st visit Testing for probab5iltyof being a carrier(o.0rselves

2.ndvisit R~i~i v, ~ ~ vi~i, Pgvvdvst Result to counse lee |- courmielle P-sregnacy confimationaiid counseilig moptas ~Ibiv e,~vf~-~ nc,.ii,vst5.(ourseNes) tobe prgnvvn

Cousele expreses anin1e t in armnocenlesisfor fetal seX deterrmnation

3rd visit

iQbstetricio. cytvg.v

/ ~~~~~~ourselvesI

4.th visit |Con|in Amones|Smvg|yoeeit cyta9enetrsbbn|

5th visit It fus is male _i

vvnestpy counsellee.|-

vvicssion of risks Cvii.lee decidl Are nge, s modevnd benefits nve fetosci py _ ith fetoscopistfobskra)_ elsewhere _

(obstetrician6th visit If fetus*svRu py

anI U = gven too vvvileelto have (iiena c

cytogeneticist.llocal physicion

|Cournselling obout ondhor|abortion (obstetrician.

Icytogeneticist

FIGURE Algorithm showing the sequence of actionstaken by a pregnant woman at high risk forhaemophilia who uses the panoply of consultantsand services at Chapel Hill. Counsellee's actionsare contained in open brackets. The actions of themedical consultants are enclosed in rectangles. Theterm 'ourselves' within certain rectangles refers toone or more of the authors of this report. 'CRM' isshorthandfor 'cross-reacting material', in thisinstance the factor VIII coagulant antigen(VIII:CAg). The only haemophilic malesdiagnosable in utero by fetoscopy are those fromkindreds in which affected makes have low toabsent VIII:CAg.

27



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/

28Emily S Barrow, Connie H Miller, HowardM Reisner, and John B Graham

counselling algorithm we follow with women whoare not pregnant was described earlier.22 The moreelaborate procedure required for those who arepregnant is shown in the figure. It indicates the rolesthat are played by ourselves, an obstetrician, acytogeneticist, and a fetoscopist and his associates.The procedure is complex, labour intensive, andexpensive, the pregnant counsellee being offeredcounsel at least eight times during six visits. It is ourcounselling policy to be 'non-directive', assisting thecounsellee to make her own decisions at each stepwith the best information and understanding that wecan muster.

GENETIC AND LABORATORY PROCEDURESPedigree information, including the previous re-productive performance of a counsellee and herancestors, and any usable linkage information, arecollected and analysed as described elsewhere.1621These data provide the 'pedigree (or prior)probability' of heterozygosity (carriership) (ito)which will be combined with information from thelaboratory to provide the 'final probability' (P(C)).

Citrated plasma is prepared and stored as describedelsewhere.23 Assays for VIII:C and VIIIR:Ag arecarried out as described earlier,23 the controlstandard being successive and sequentially calibratedpools of normal plasma stored at -700C. The datafrom the laboratory together with age are used tocalculate a 'likelihood ratio' (LR), favouringcarriership by linear discriminant analysis asdescribed by Elston et al.*16 The LR, which is theratio between the likelihood of being a carrier andthe likelihood of being normal, is combined math-ematically with ixc to produce the final probability ofcarriership (P(C)) which ranges from 1 *0 (certain) to0 0 (impossible). We have programmed the completestatistical procedure for a desktop computer(HP 9815A), and it requires only 15 seconds toobtain P(C) once irc, age, and the VIII :C andVIIIR:Ag values are available.

Results

OBLIGATORY CARRIERSCertain women in haemophilia families arerecognised genetically as heterozygous and arereferred to as obligatory (or obligate) carriers. Thedaughters of men with X linked haemophilia(paternal carriers) are so regarded because the Xchromosome which they receive from their fatherscarries the haemophilia gene. Women with two ormore haemophilic sons, or an affected son and a

*The discriminant is: [(-0.3746-0.0496 a) x + (4.447-0055 a)y],where x = In VIIIR:Ag, y =In VIII:C, and a = age in years. It isnormally distributed in carriers with a mean of (16.16-0-2363 ac) anda variance of 3.09. It is normally distributed in normal women with amean of (20 - 52-0* 269 a) and a variance of I * 11.

carrier daughter, are so regarded, since the prob-ability is negligible that mutation will have occurredtwice in their matings. Also, a woman in the main-line of descent of a haemophilia family who has asingle affected son or a carrier daughter is regardedas an obligatory carrier. We refer to all three of thelatter types as 'maternal carriers'. There were 36obligatory carriers among our 194 non-pregnantcounsellees (15 maternal and 21 paternal) and thedata concerning them are summarised in table 1.The table has been divided into two sections to

display the data separately for maternal carriers (A)and paternal carriers (B) because it has been reportedthat the VIII :C levels of paternal carriers aresignificantly lower than those ofmaternal carriers,24 25an assertion which has been vigorously disputed byothers.26-28 The women within each section have beenfurther arranged by age and family. The 15 maternalcarriers can be seen to have been unrelated, while the21 paternal carriers fell into 16 families.t Individualcoagulation data are shown as well as individuallikelihood ratios and final probabilities.

There was no significant regression of VIII:C orVIIIR:Ag on age in either maternal carriers (n = 15)or paternal carriers (n = 21 or n = 16), and it isapparent on inspection that there is no significantdifference between maternal and paternal carriers inrespect of VIII:C or VIIIR:Ag levels. When thevalues for all the members of a group of relatedpaternal carriers are averaged and the averageentered as a single value (n = 16), there is a slightbut not significant decrease in the mean levels ofVIII:C and VIIIR:Ag.

Table 1 also displays the LRs and two estimates ofP(C) for each obligatory carrier. An LR greater than1:1 means that carriership is likely, and a value lessthan 1:1 means that carriership is unlikely. Thetable shows that LRs greater than 1:1 werepresent in 30/36 (83 %) of obligatory carriers, whilevalues of less than 1:1 were present in 6/36 (17%).The aberrant LRs presumably represent the effects oflyonisation and provide an estimate of the rate offalse negative diagnosis caused by it.Two estimates of final probability (P(C)) are given.

The correct one is that which is based on the correctpedigree probability for obligatory carriers(Tcc = 1 0). This means that P(C) for obligatorycarriers must also be 1-0, regardless of the LR.When, however, the women are treated as thoughthey were sisters of haemophiliacs (xCc = 0 50) it isseen that the six women with LRs less than 1:1 haveP(C)s less than 0 50 and would have been diagnosedas non-carriers on the basis of laboratory data alone.Maternal carrier 9 is particularly interesting. Her

tMaternal carrier 9 and paternal carrier 15 were 4th cousins.

28



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/


TABLE 1 Obligatory carriersLikelihood ratio Final probability P(C)

Family Subject Age VIII:C VIIIR:Ag(t/dl) (p/dl) >1:1 <1:1 7rC= 1-0 rc = 0-50

A Maternal carriersI 1 25 34 63 9 x 104:1 t >0-999II 2 26 26 125 5x109:1 >0.999III 3 30 28 75 5 x 106:1 >0.999IV 4 30 70 79 7:1 0*867V 5 32 60 51 4:1 0-783VI 6 34 66 109 54:1 0*982VII 7 36 83 129 10:1 0.910VIII 8 36 90 188 35:1 1.0 0.972*IX 9 39 167 132 1:11 0*086tX 10 39 76 111 11:1 0.914XI 1 1 42 52 94 263:1 0.996XII 12 48 75 87 2:1 0.676XIII 13 49 76 119 13:1 0-929XIV 14 50 147 149 1:3 0*236XV 15 58 45 91 332:1 0-997

R=38-3 73 106-8SD= 9-6 39-6 35.5

Range 25-58 26-167 51-149

B Paternal carriersXVI 1 3 23 92 6x 1011:1 t >0 999XVII 2 4 60 74 332:1 0*997XVIIIa 3 10 57) 91 991:11 0.999b 4 13 61 59 113 126 999:1 t 0.999

c L 5 20 59J 154 1000:1 J >0-999XIX a f 6 11 135k90 5 230 174 1:42} 0.192

b 7 16 46 118 5 x104:1 j>0.999XX a 8 12 72 195 70 131: -1b { 9 14 167}19.5 156 113 11:1 1:25} 1.0 0-918 0.038XXI 10 16 51 116 8 x 103:1 >0.999XXII 11 18 57 94 499:1 0-998XXIII 12 18 46 137 1 xlO5:1 >0.999XXIV 13 19 58 126 1-3 x 103:1 >0.999XXV 14 20 64 72 26:1 0.963*IX 15 20 93 240 28:1 0.966XXVI 16 21 45 62 999:1 0.999

b 18 25 86} 81 1-22:1 10.134

$XXVIII 19 23 44/33 107/90 4 x 104:1 >0-999$XXIX 20 28 23/21 79/185 8 x 108:1 >0-999$XXX 21 35 184/149 157/129 1:13/1:8 0-072/0-108

C x=17 5n=21 SD= 7-5

t Range = 3-35f x=18.1n=16 SD= 7.9

72-742-421-18469-240.0

116-145.950-230114-546-8

LR was only 1:11, but she produced three affectedsons.There is no evidence in table 1 which supports a

familial tendency towards similar likelihood ratios.There are five sets of related obligatory carriers:families IX, XVIII, XIX, XX, and XXVII. Family IXconsists of maternal carrier 9 and paternal carrier 15.The former has an LR of 1 :11 and the latter has anLR of 26:1. Family XVIII consists of three daughtersof a haemophiliac with LRs >999:1. In familiesXIX, XX, and XXVII, each consisting of a pair ofdaughters of a haemophiliac, one sister has an LR>1 :1 and the other has an LR <1:1.

POTENTIAL CARRIERSWhen a female relative of a haemophiliac is not an

*Belong to same family.tHad 3 affected sons.tTested on 2 separate occasions.

obligatory carrier, she must be regarded as apotential carrier unless she shows herself to beindubitably heterozygous by producing an affectedson or a carrier daughter. Until heterozygosity hasbeen proved by transmission of the haemophiliagene, however, it is possible only to assign her aprobability of carriership. Table 2 displays the finalprobabilities obtained on 158 potential carriers. TheP(C)s have been grouped by quartiles, 1-00-0 75being the quartile most likely to be carriers, and0-24-040 being the quartile least likely to be carriers.

Seventy-eight sisters of haemophiliacs are shownin row (a), 62 second and third degree relatives inrow (b), and 18 mothers of sporadic haemophiliacsin row (c). In each row, the majority of values are inthe first and fourth quartiles, 73-74/158 (47 %) in the

29



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/

Emily S Barrow, Connie H Miller, HowardM Reisner, and John B Graham

TABLE 2 Potential carriers, not pregnant

Final probabilities P(C)

Carriers Not carriersNo When 1 + 2 3 + 4

7rc is: (1 8-075) (0.74-S50) (0-49-0-25) (0-24-00)

(a) Sisters of haemophiliacs 78 0-50 38 8 9 23

46 32(b) Second and third degree relatives 62 0-36* 20 7 6 29

27 35(c) Mothers of sporadic haemophiliacs (1) 18 0.67 15 1 1 1

16 2(2) 18 0-83 16 1 0 1

17 1Total 158 73-74 16 16-15 53

*Average of 62 individual pedigree probabilities ranging from 0 * 01-0 50.(a) Expected = 50:50 (39:30); observed = 46:32. x2 = 2-5, p>0- 10(b) Expected = 36:64 (22:40); observed = 27-35. x2 = 1*76, p>0. 10(c) (1) Expected = (12:6); observed = (16:2). x2 = 3.999, p<0.05

(2) Expected = (15-7:2.3); observed = (17.1). X2 = 0-8423, p>0-30

largest quartile, and 53/158 (33%) in the smallestquartile. Thus, the probability that a potentialcarrier was indeed a carrier was either very great orvery slight in 80% of instances. Classification byfinal probabilities also provided a means of com-paring the congruence of the assigned genotype(P(C)) with the Mendelian expectation (7tc). This wasaccomplished by combining the totals in the twolargest quartiles (1* 0-050) and the two smallest(0A49-00) in each subset of counsellees to formtestable ratios of carriers: normals.

Since sisters of haemophiliacs, whose mothers arecarriers, have a 1:1 chance of being carriers, the 78sisters of haemophiliacs in row (b) are expected todivide equally (39:39) between carriers and normals.The 46:32 ratio observed between the high and lowprobability groups might have occurred more than10% of the time by chance alone.The prior probability for the heterogeneous set of

62 second and third degree relatives was calculatedby averaging the prior probabilities of the individualwomen, which ranged from 0 01 to 0 50. The result0 36 implied that a 22:40 ratio of carriers:normalswas expected among the 62 subjects. The observation27:35 was well within expectation.

Mothers of sporadic haemophiliacs posed aproblem because the selective disadvantage ofhaemophiliacs has not been determined since thebeginning of the modern era of blood coagulationresearch. Haematologists are aware, however, thatmany severe haemophiliacs are reproducing. Wedecided, therefore, to examine the mothers ofsporadic haemophiliacs under two assumptions:(1) that the trait is lethal in affected males ( Tc = 0 67)

and (2) that fitness in haemophilic males is 50% ofnormal, that is, xtc = 0 83*.

Line I of row (c) shows that a significant excess ofmothers scored as carriers under the assumption oflethality.

Line 2 of row (c) shows that the assumption of50% fitness brings expectation and observationalmost into balance.

COUNSELLEES WHO WERE PREGNANTWHEN FIRST SEENDuring the period of study 20/214 (9%) of thecounsellees were pregnant when first seen. Seventeenwere potential and three were obligatory carriers.Pertinent data are summarised in table 3.

Thirteen potential carriers had P(C)s between 0 * 02and 0 33 and elected to go through pregnancy

TABLE 3 Potential carriers who were pregnant on thefirst visit before fetoscopy was available.No P(C)s Fetuses Amnio- Abortions Livebornof at risk centeses non-haemophilicmothers children

Male Female

13 0.02- 13* 0 0 4 70-33

1 0-50 lt 1 0 0 01 0-83 1 1 0 0 11 >0.99 1 0 0 1 01 >0.99 3 2 1 0 23 1.00 3 3 1 0 2

20 - 22 7 2 5 12

*One lost to follow-up, one pendingtPending, male

*The theory which produces this estimate assumes genetic equilibriumand that mutation rates are equal in ova and sperm.29

30



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/


without further study. They delivered 11 non-haemophilic children, two not yet having beenaccounted for. Three of four mothers with P(C)sbetween 0 50 and 1 - 00, with six fetuses at risk in fivepregnancies, requested amniocentesis. The fouramniocenteses carried out identified three femalesand two males. One male was aborted, one went toterm and was normal at birth, and one male birth ispending. All three females were non-haemophilic atbirth. Three obligatory carriers requested amnio-centesis and the single male which was identified wasaborted.

AMNIOCENTESIS AND FETOSCOPY

It has been possible for many years to determine thesex of a fetus in utero by amniocentesis. Recently ithas become possible to determine the haemophiliastatus of most male fetuses by fetoscopy andexamination of fetal blood.30 31 The procedure isdifficult, poses a significant risk to the fetus, and isusually restricted to high-risk mothers.

Since 1977, seven of the 158 potential carriersincluded in table 2, who came from families of severeCRM- haemophilia and had been seen beforebecoming pregnant, have returned for counsel afterbecoming pregnant. Each requested amniocentesisand those with male fetuses requested fetoscopy.This experience is summarised in table 4.A female fetus (4) identified by amniocentesis was

carried to term and is non-haemophilic. A maleidentified before fetsocopy became available (3a) wasaborted but was not necropsied. Six males identifiedby amniocentesis were fetoscoped. One could not bediagnosed but was not haemophilic at birth. Fourwere diagnosed as normal at fetoscopy and three ofthese were normal at birth, the fourth not survivingthe procedure. The only male diagnosed as haemo-philic by fetoscopy did not survive the procedure,necropsy confirming the diagnosis.

Discussion

There is general agreement that the optimal pro-cedure when counselling a woman potentiallyheterozygous for an X linked disorder is to establishher probable genotype as precisely as possible, pointout her options, and let her decide what she wantsdone. When, as in haemophilia, there is not aqualitative test which distinguishes heterozygouscarriers from normal women with certainty, and thequantitative tests produce overlapping ranges ofvalues in reference groups, probability methods mustbe used. Our procedure is to combine the likelihoodratio of carriership obtained from laboratory databy linear discriminant analysis with the priorprobability drawn from pedigree information toyield the final probability of carriership (P(C)). P(C)is the datum which is communicated to the counselleeand whose implications are explained to her.Both counsellor and counsellee must appreciate,

however, that no matter how high or low the P(C), itis only a probability, not a certainty! A pertinentexample is seen in table 2. Had subject 9 of family IXvisited us before having had children, a priorprobability of 0 50 would have been assigned and,together with her laboratory data, would haveyielded a P(C) of 0-086. This low value might haveled her to believe that she was normal. In fact, sheproduced three affected sons in succession, provingthat her correct Trc and P(C) are 1 * 0. Her phenotypeand history are reminiscent of an obligatory carrierdescribed earlier.32The reader might wonder why the mean VIII:C

level of our obligatory carriers was 72 85 ,u/dl whenlyonisation theory suggests that it should have been50. There are several possible explanations. Theassays are not perfect. There are two types ofprocedure which do not always give the same resultsand the coefficients of variation are sizeable.

TABLE 4 Women with high P(C)s who returned after becoming pregnanltInitial visit Subsequent visit

Sex offetus Phenotypeat

Year Ttc P(C) Year amniocentesis At fetoscopy At birth At necropsy

(1) 1975 0-5 0-94 1979 Male (Unsuccessful) Premature, normal(2) 1975 0.25 0.98 1979 Male Normal Normal(3) 1976 0.5 >0-99 (a) 1978 Male * (Aborted) Not done

(b) 1979 Male Haemophilic --t Haemophilic(4) 1976 0-25 >0-99 1979 Female Not haemophilic(5) 1976 0*5 0*65 1979 Male Normal Normal(6) 1980 0.50 0°50$ 1980 Male Normal t Normal(7) 1980 0-50 0.86 1980 Male Normal Normal

*Fetoscopy was not available at the time.tDid not survive fetoscopy.:Fetoscopy at mother's insistence.

31



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/


Furthermore each laboratory produces its own localstandard for use as control. The usual practice,which we follow, is carefully to prepare plasmas froma number of presumably normal persons, then poolthem and store aliquots at low temperature tominimise activation and maximise stability. As doothers, we assume that the VIII:C value of eachfreshly thawed aliquot is 100 ,u/dl. Since poolsoriginate from a restricted number of subjects andthe number ofobligatory carriers is limited, stochasticprocesses may act on either. A significant stochasticeffect on our material cannot be assessed, however,because a suitable universal standard is lacking.Published reports suggest that our experience is notunique, since the VIII:C level of the obligatorycarriers was 78 7 ± 46:6 in the classical study ofZimmerman et a16 and 78 ± 41 in the study ofEkert et al.33An issue vigorously debated at present is whether

obligatory carriers of paternal and maternal origindiffer, specifically whether paternal carriers havelower VIII:C levels than maternal carriers. This isnot simply a technical disagreement among coagu-lationists or a theoretical matter of concern only togeneticists interested in inactivation of the Xchromosome. It also has practical significance forcarrier detection. The question being asked in carrierdetection is: did this woman receive a haemophiliagene from her mother? Or in our usage: is she amaternal carrier? If there is a significant differencebetween paternal and maternal carriers, paternalcarriers should be excluded from reference groups ofobligatory carriers used for classifying potentialcarriers. Our sample clearly shows no difference, yetthose of others do.24 25 Resolution of this disagree-ment will probably require the acquisition of aconsiderably larger body of data collected understandardised conditions.Those concerned with carrier detection must

remember that subsequent events may change the7cc initially assigned to a potential carrier. Forexample, a woman with a icc of 0 50, because she isthe sister of a haemophiliac, would progressivelyreduce it to 0 33 then to 0 20 and finally to 0 11 byhaving three normal sons in succession. Yet shemight have an affected son on the fourth pregnancydemonstrating that she had been a carrier all along.Furthermore, the outcome of additional randomprocesses, including the vagaries of inactivation ofthe X chromosome (lyonisation) in heterozygousfemales, has major effects on the plasma levels ofVIII:C, which in turn affect the likelihood ratio andthrough it P(C). These considerations imply that afew women among a group of carriers are expectedto have low P(C)s but to produce affected sons. Ourdata on obligatory carriers showing that 17% score

as 'normal' suggests that 17% of the true carrierswithin the larger group of potential carriers willscore as normal. We are, therefore, exceedinglycareful to explain the full meaning of a low prob-ability to a potential carrier, and always put it inwriting!We were pleased to find that the P(C)s of 80% of

potential carriers fell into the highest and lowestquartiles. This high rate of determinancy means thatthe test we have been using is relatively efficient. Weexpected 'extreme lyonisation', which tends toconceal carriers with low P(C)s while exposing thosewith high P(C)s, to produce more negative diagnosticerrors (carriers classified as normal) than positiveerrors (normals classified as carriers). In fact, theratios in rows (b) and (c) of table 3, which are notsignificantly different from expectation, suggest thatthe types of error balanced each other fairly well.There was a paucity of low scores (normal

phenotypes) among the mothers of sporadic haemo-philiacs, and the single apparently normal case mayrepresent an instance of 'extreme lyonisation'. Thefinding suggests that either the fitness of haemo-philiacs is much higher than suspected or the rate ofmutation may not be equal in sperm and ova. Therarity of normal phenotypes among the mothers ofsporadic haemophiliacs agrees with an earlier studyof sporadic haemophiliacs24 and with the findings inDuchenne muscular dystrophy34 and the Lesch-Nyhan syndrome.35 The combined experience withX linked traits suggests that most mothers ofsporadiccases are in fact carriers and that mutations are veryrarely encountered. We currently use 0 80 as the xcfor mothers of sporadic haemophiliacs and regardthis as conservative.

It was probably improper to apply our discrimi-nants to pregnant women, since the women in thereference groups on which the discriminants arebased were not pregnant. However, no one has yetprovided pregnancy based discriminants, and it willbe difficult to obtain them, because this requiresprospective study of adequate numbers of pregnantcarriers. Our pregnant counsellees were either in the1st or early 2nd trimester of pregnancy, however, andwe seem to have been able to identify them correctly,at least P(C)s ranged from 0 02 to >0 99.We wondered why the majority of women at risk

for having haemophilic sons do not use the opor-tunities to become better informed. Our catchmentarea contains approximately 600 haemophiliacs andtherefore approximately 3600 possible carriers.'8 Yetonly 214 counsellees have come to us for advice. Ofcourse, many women in the population are pre- orpost-menopausal, but as many as 1000 may be nubile.A survey by Evans and Shaw in the UK 36 showed

that most obligatory carriers of haemophilia prefer

32



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/


either to avoid pregnancy or to risk its consequencesrather than undergo amniocentesis and fetoscopy.The British survey and our experience may becongruent, and most women in both countries maybe expressing the view that, since the health caresystem cannot guarantee a perfect child, 3 thebenefit of obtaining a non-haemophilic son is notsufficient to balance the psychological trauma ofhaving to make a series of agonising decisions.However, we do not know how many of the 786(1000-214) potential nubile carriers who have notconsulted us have been pregnant, nor how many ofthe 194 studied have become pregnant since seeingus. But seven potential carriers, who were seen

before becoming pregnant, have returned for furtheradvice after becoming pregnant, and 20 potentialcarriers whom we had never seen sought aid afterbecoming pregnant. Thus our counsellees may bereacting differently from those encountered by Evansand Shaw.36 Some of the women in the United Statesat high risk for haemophilic sons appear to bewilling to use available technology, and this may

indicate that educational efforts are slowly beginningto bear fruit. On the other hand it may merelyindicate that the population of potential mothers ofhaemophiliacs is dichotomous in America, one

subset being willing to use the aid available, theother preferring to leave the result to chance.

The authors wish to thank Dr Harold Roberts andDr Philip Blatt of the UNC Comprehensive Hemo-philia Diagnostic and Treatment Center for referralof many of the counsellees.

References

Morgan TH. Sex limited inheritance in Drosophila.Science 1910;32:120-2.

2 Macklin MT. Heredity in hemophilia. Am J Med Sci1928;32:218-24.

3 Veltkamp JJ, Drion EF, Loeliger EA. Detection of thecarrier state in hereditary coagulation disorders. II.

Thromb Haemostas 1968;19:403-22.4 Bennett E, Huehns ER. Immunological differentiation ofthree types of haemophilia and identification of somefemale carriers. Lancet 1970 ;ii :956-8.Zimmerman TS, Ratnoff OD, Powell AE. Immunologicdifferentiation ofclassic hemophilia (factor VIII deficiency)and von Willebrand's disease: with observations oncombined deficiencies of antihemophilic factor andproaccelerin (factor V) and an acquired circulatinganticoagulant against anti-hemophilic factor. J Clin Invest1971 ;50:244-54.

6 Zimmerman TS, Ratnoff OD, Littell AS. Detection ofcarriers of classic hemophilia using an immunologic assayfor anti-hemophilic factor (factor VIII). J Clin Invest 1971;50 :255-8.

7 Bennett B, Ratnoff OD. Detection of the carrier state forclassic hemophilia. N Engl J Med 1973 ;228 :342-5.

8 Denson KWE. The detection of F VIII-like antigen inhaemophilic carriers and in patients with raised levels ofbiologically active factor VIII. Br J Haematol 1973 ;24:451-61.

9Ekert H, Helliger H, Muntz RH. Detection of carriers ofhaemophilia. Thromb Haemostas 1973 ;30:255-62.

20 Hoyer LW, Rick ME. Implications of immunologicmethods for measuring anti-hemophilic factor (factorVIII). Ann NYAcad Sci 1975;240:362-9.

1 Bouma BN, van der Klaauw MM, Veltkamp JJ,Starkenburg AE, van Tilburg NH, Hermans J. Evaluationof the detection rate of hemophilia carriers. Thromb Res1975 ;7:339-50.

12 Graham JB, Barrow ES, Elston RC. Lyonization inhemophilia: a cause of error in direct detection ofheterozygous carriers. Proc NY Acad Sci 1975;240:141-6.

13 Meyer D, Plas A, Allain JP, Sitar JP, Larrieu MJ.Problems in the detection of carriers of haemophilia A.J Clin Pathol 1975 ;28 :690-5.

14 Prentice CRM, Forbes CD, Morrice S, McLaren AD.Calculations of predictive odds for possible carriers ofhaemophilia. Thromb Haemostas 1975 ;34:740-7.

`5 Rizza CR, Rhymes IL, Austen DEG, Kernoff PBA,Aroni SA. Detection of carriers of haemophilia: a 'blind'study. Br J Haematol 1975;30:447-56.

16 Elston RC, Graham JB, Miller CH, Reisner HM,Bouma BN. Probabilistic classification of hemophilia Acarriers by discriminant analysis. Thromb Res1976;8 :683-95.

17 Klein HG, Aldedort LM, Bouma BN, Hoyer LW'Zimmerman TS, DeMets DL. A cooperative study for thedetection of the carrier state of classic hemophilia. N EngiJ Med 1977;296:959-62.

18 Akhmeteli MA, Aledort LM, Alexaniants S, et al.Methods for the detection of haemophilia carriers: amemorandum. Bull WHO 1977 ;55 :675-702.

19 Ratnoff OD, Jones PK. The laboratory diagnosis of thecarrier state for classic hemophilia. Ann Intern Med 1977;86:521-8.

20 Graham JB, Barrow ES, Flyer P, Dawson DJ, Elston RC.Identifying carriers of mild haemophilia. Br J Haematol1980;44:671-8.

21 Graham JB. Genotype assignment (carrier detection) inthe haemophilias. Clin Haematol 1979 ;8:115-45.

22 Graham JB. Population genetics and genetic counselingin hemophilia A (classic hemophilia). Proc XIth CongWorld Fed Hemophilia 1976:1-10.

23 Reisner HM, Katz HJ, Goldin LR, Barrow ES,Graham JB. Use of a simple visual assay of Willebrandfactor for diagnosis and carrier identification. Br JHaematol 1978 ;40:339-50.

2' Biggs R, Rizza CR. The sporadic case of haemophilia A.Lancet 1976;ii:431-3.

25 Chediak J, Telfer MC, Jaojaroenkerl T, Green D. Lowfactor VIII coagulant activity in daughters of subjectswith hemophilia A compared to other obligate carriers.Blood 1980;55 :552-8.

26 Kasper CK. Letter to the Editor. Blood 1980;56:742.27 Graham JB. Letter to the Editor. Blood 1980;56:742.28 Lusher JM, Warrier Al, Khalifer AS. Letter to the

Editor. Blood 1980;56:1153.29 Morton NE, Lalouel JM. Genetic counseling in sex

linkage. Birth Defects 1979;15:9-24.30 Firschein SI, Hoyer LW, Lazarchick J, et al. Prenatal

diagnosis of classic hemophilia. N Engl J Med 1979 ;300:937-41.

31 Mibashan RS, Thumpston JK, Singer JD, et al. Plasmaassay of fetal factors VIII :C and IX for prenatal diagnosisof haemophilia. Lancet 1979;i:1309.

33



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/


32 Graham JB, Miller CH, Reisner HM, Elston RC,Olive JA. The phenotypic range of hemophilia A carriers.Am J Hum Genet 1976;28:482-8.

33 Ekert H, Helliger H, Muntz RH. Detection of carriers ofhaemophilia. Thromb Haemostas 1973 ;30:255-62.

34 Roses AD, Roses MJ, Miller SE, Hull KL, Appel SH.Carrier detection in Duchenne muscular dystrophy.N Engl J Med 1976;294:193-8.

35 Francke U, Felsenstein J, Gartler SM, et al. The occur-rence of new mutants in the X-linked recessive Lesch-Nyhan disease. Am J Hum Genet 1976; 28:123-7.

36 Evans DIK, Shaw A. Attitudes of haemophilia carriers tofetoscopy and amniocentesis. Lancet 1979;ii:1371.

37 Karp LE. Editorial column: Toward the perfect child.Am J Med Genet 1980;5:1 15-6.

Requests for reprints to Professor J B Graham,Pathology Department, Building 228H, University ofNorth Carolina, Chapel Hill, North Carolina27514, USA.

34



g.bmj.com

/J M


g.19.1.26 on 1 February 1982. D

ownloaded from

http://jmg.bmj.com/

genetic counsellingprematurely, andonehaemophilic maleandonenormalmaledid not survive fetoscopy....

Documents